Combining a Pharmacological Network Model with a Bayesian Signal Detection Algorithm to Improve the Detection of Adverse Drug Events

Abstract

Introduction: Improving adverse drug event (ADE) detection is important for post-marketing drug safety surveillance. Existing statistical approaches can be further optimized owing to their high efficiency and low cost.Objective: The objective of this study was to evaluate the proposed approach for use in pharmacovigilance, the early detection of potential ADEs, and the improvement of drug safety.Methods: We developed a novel integrated approach, the Bayesian signal detection algorithm, based on the pharmacological network model (ICPNM) using the FDA Adverse Event Reporting System (FAERS) data published from 2004 to 2009 and from 2014 to 2019Q2, PubChem, and DrugBank database. First, we used a pharmacological network model to generate the probabilities for drug-ADE associations, which comprised the proper prior information component (IC). We then defined the probability of the propensity score adjustment based on a logistic regression model to control for the confounding bias. Finally, we chose the Side Effect Resource (SIDER) and the Observational Medical Outcomes Partnership (OMOP) data to evaluate the detection performance and robustness of the ICPNM compared with the statistical approaches [disproportionality analysis (DPA)] by using the area under the receiver operator characteristics curve (AUC) and Youden’s index.Results: Of the statistical approaches implemented, the ICPNM showed the best performance (AUC, 0.8291; Youden’s index, 0.5836). Meanwhile, the AUCs of the IC, EBGM, ROR, and PRR were 0.7343, 0.7231, 0.6828, and 0.6721, respectively.Conclusion: The proposed ICPNM combined the strengths of the pharmacological network model and the Bayesian signal detection algorithm and performed better in detecting true drug-ADE associations. It also detected newer ADE signals than a DPA and may be complementary to the existing statistical approaches.

Adverse drug reaction signal detection methods in spontaneous reporting system: A systematic review

“…A total of 4197 literatures were identified through initial search. After selection, 93 literatures 4–8,11–98 met our inclusion criteria and were included, covering a period from 1998 to 2022. The literature screening process is shown in Figure 1.…”

“…The 93 included literatures contained 27 reviews 19,21,22,30,34,35,40–42,44,47,50,55–57,61,63,67,68,73,77,82,83,85,92,94,98 and 66 methodological studies 4–8,11–18,20,23–29,31–33,36–39,43,45,46,48,49,51–54,58–60,62,64–66,69–72,74–76,78–81,84,86–91,93,95–97 . Among them, 51 were from China, 18 were from America, 6 were from England, and 5 were from France.…”

“…A series of studies have compared the performance of different signal detection methods 4,[11][12][13][14][15] However, these previous studies have two notable limitations and are unable to provide effective guidance for the optimum selection of signal detection method. First, each of these previous studies only compared a small number of signal detection methods, and thus they were unable to summarize the characteristics and application scopes of all available signal detection methods.…”

An Early Adverse Drug Event Detection Approach with False Discovery Rate Control

“…22 Recent studies utilized drug similarity network to enhance existing SDAs. [23][24][25] For instance, Liu and Zhang used drug similarity network to enhance signals generated from PRR, ROR, GPS and BCPNN. 24 Ji et al enhanced signals from BCPNN by leveraging drug similarity network derived prior ADE risk distribution.…”

“…24 Ji et al enhanced signals from BCPNN by leveraging drug similarity network derived prior ADE risk distribution. 23 While enhanced ADE signals demonstrated improved performance, FDR control remains unclear on prioritizing these signals. Additionally, the performance of early ADE detection has not been explicitly investigated under a time-to-event setting (e.g., time from first report to time of ADE detection).…”